Abstract：Volatile Organic Compounds (VOCs) are not only present in the human living conditions, but also can be produced in the process of human metabolism. The types and kurtosis of VOCs produced by the human body are closely related to human health and disease status. In recent years, analyzing VOCs components and concentration in human exhaled breath through non-invasive detection can be used to indicate human health status, which has been widely concerned in the field of clinical diagnosis and health monitoring. This paper reviews the clinical application value and common detection methods of exhaled VOCs, and elucidates pros and cons of each method . In this paper, the revolutionary progress brought by the new electronic nose technology to the detection of exhaled VOCs is described, which brings more possibilities for the high sensitivity, high throughput and intelligent detection of exhaled VOCs.

Abstract： Resonant microcantilever is a mass-type sensor that transforms mass change into frequency-shift signals. Due to the advantages of high mass resolution, high sensitivity, low cost, easy integration and miniaturization, resonant microcantilevers have received extensive attention from researchers. In recent years, with the rapid development of MEMS (Micro-Electro-Mechanical Systems) technology, resonant micro-cantilever sensors have been widely used in mass and flow control, trace biomedical detection, gas or liquid molecular analysis and other broad application fields. Herein, the specific applications of resonant microcantilever sensors in environmental detection, biomedicine and other fields in recent years are reviewed, and the future development trends are prospected.

Abstract：Abstract： The assessment of coagulation can help diagnose blood diseases, guide treatment of cardiovascular diseases, and assess risk of bleeding during surgery. Physical coagulation testing is of great interest to clinical medicine because of its short detection time, high sensitivity, and simple operation. With the rapid development of optical coherence tomography (OCT) technology, it has been used in the blood coagulation detection. In this paper, we summarize the physical methods for the blood coagulation detection in recent years and specially review the applications of the OCT methods in the coagulation testing. The future development of blood coagulation detection is also prospected.

Abstract：Abstract： Cell traction force microscopy method is one of the mainstream tools for measuring cell traction force, which uses fluorescent microbeads as markers to measure the deformation of the substrate and calculate cell traction force. However, due to the random distribution of fluorescent microbeads in the substrate, the calculation accuracy of the cell traction force field maybe affected. To address the problem of uncontrollable depth of fluorescent microbeads, a method using patterned elastic substrate was proposed to measure cell traction forces, in which the micropattern arrays fabricated on the surface of substrate was used as deformation markers. A micro-pad array with diameter of 3 µm and height of 0.8 µm was designed and fabricated on the polydimethylsiloxane substrate surface as the displacement markers. In the finite element simulation, external forces mimicking cell traction forces was applied at different locations on the substrate surface and the displacements field of the micro-pad array was obtained. Using the displacements field obtained by simulation, the traction force field was calculated through the traction force inversion algorithm, and the results were consistent with the input external forces in the simulation. The patterned elastic substrates were validated by mapping neonatal rat ventricular myocytes contraction forces. Both simulation and experiment demonstrate that the capacity of patterned elastic substrate for generating accurate cell contraction force maps, providing new research method for pathological study of myocardial diseases.

Abstract：Abstract： A multi-channel real-time system is designed for monitoring photoplethysmographic (PPG) using RISC-V kernel. The system takes E203 IP of RISC-V architecture as the low-power core, which connects analog front-end module and Bluetooth control module through ICB bus. The analog front-end control module controls back-to-back LEDs to realize two-channel PPG signal acquisition. Multi-channel acquisition can be realized by reusing the analog front-end control module. The Bluetooth control module is connected to E203 to deliver system control commands, and then upload PPG data to the terminal. Finally, The FPGA system verifies the 22-bit dual-channel PPG signal Internet of Things terminal application.

Abstract：Abstract： Computer text analysis is an important branch in the field of natural language processing, and it is a computer technology that studies how to extract various types of information from a given corpus from text data. At present, computer text analysis has entered a new historical stage. On the one hand, the keyword extraction algorithm has gradually been completed. On the other hand, with the emergence of the BERT method, the word vector calculation problem has also made great progress. However, there are still some problems to be solved in both keyword extraction and word vector calculation. In addition, many existing studies suitable for using text analysis still use ancient text analysis methods. Therefore, in the future, how to better reduce the model size to promote the integration of disciplines and improve the comprehensive social benefits of text analysis will become an important issue in the development of text analysis algorithms.

Abstract：Abstract： Deep neural networks have achieved great success in classification and prediction of high-dimensional data. Training deep neural networks is a data-intensive task, which needs to collect large-scale data from multiple data sources. These data usually contain sensitive information, which makes the training process of convolutional neural networks easy to leak data privacy. Aiming at the problems of data privacy and communication cost in the training process, this paper proposes a distributed training method for deep neural networks, which allows to jointly learn deep neural networks based on multiple data sources. Firstly, a distributed training architecture is proposed, which is composed of one computing center and multiple agents. Secondly, a distributed training algorithm based on multiple data sources is proposed, which allows to distributed jointly train models through the splitting of convolutional neural networks under the constraints that raw data are not shared directly and the communication cost is reduced. Thirdly, the correctness of the algorithm is analyzed. Finally, the experimental results show that our method is effective.

Abstract：Abstract： In view of the complex structure and high maintenance cost of transformers, this paper proposes a transformer fault signal recognition algorithm based on deep learning. Firstly, the voiceprint signal under the working condition of the transformer is analyzed and the two-dimensional image signal is converted. Based on the advantages of VGG16 neural network in the image, a MCA attention mechanism is proposed, which can retain both background information and detail information. Secondly it optimizes the maximum pooled down sampling in VGG16, and adopts a soft pooled sampling method to reduce the feature loss caused by the maximum pooled down sampling in the image. Finally, in order to avoid the occurrence of over fitting, the activation function in the top structure of VGG16 is optimized, and the SELU activation function that can be self normalized is quoted. The experiment proves that the generalized S-transform is the best choice for converting one-dimensional time-domain signal to two-dimensional image signal, and the average recognition rate of the proposed algorithm for six types of fault signals reaches 99.15%.

Abstract：Abstract： With the increasing resolution of the input image of the current target detection task，the feature information extracted from the feature extraction network will become more and more limited under the condition that the receptive field of the feature extraction network remains unchanged，and the information coincidence degree between adjacent feature points will also become higher and higher.This paper proposes an FSA(fusion self-attention)-FPN，and designs SAU(self-attention upsample) module.The internal structure of SAU performs cross calculation with self-attention mechanism and CNN to further Feature fusion，and reconstructs FCU(feature coupling unit) to eliminate feature dislocation between them and bridge semantic gap. In this paper，a comparative experiment is carried out on Pascal VOC2007 data set using YOLOX-Darknet 53 as the main dry network. The experimental results show that compared with the FPN of the original network，the average accuracy of MAP@ [.5:.95] after replacing FSA-FPN is improved by 1.5%，and the position of the prediction box is also more accurate.It has better application value in detection scenarios requiring higher accuracy.

Abstract：Abstract： Based on dual active feedback with miller capacitor compensation (DAFMCC), a novel single pole system of capacitor-less on-chip low-dropout (LDO) voltage regulator circuit is presented. The LDO adopts a fully differential amplifier as the first gain stage, two parallel feedforward pathways can improve the transient response performance of the circuit meanwhile insert a push-pull stage at the gate of the power MOSFET which results in a symmetrical slewing behavior. The proposed LDO regulator is verified by TSMC 0.18 µm BCD process. The simulation results show that when supply voltage changes from 2 V to 4 V, the minimum dropout voltage is 200 mV, the maximum load current is 120 mA. The transient response recovery time is smaller than 0.7 µs, the phase margin is larger than 60°.

Abstract： Aiming at the problem that the traditional SATA controller has a single interface and cannot give full play to the performance of the SSD, a multi-channel high-speed storage circuit based on PCIE to SATA is designed. This design makes full use of the high-bandwidth and low-latency characteristics of the PCIE bus and follows the AHCI protocol, which can greatly reduce the useless seek times and data search time of the hard disk, and improve the read performance. At the same time, the design can support 4 SATA channels and has good scalability. This design combines the characteristics of PCIE and SATA protocols, introduces the system architecture of PCIE to SATA high-speed storage circuit, and elaborates the data stream transmission process based on AHCI protocol. Finally, the circuit is tested based on the FPGA prototype verification. The single-disk read and write rates of the circuit are 562MB/s and 527MB/s, respectively. Compared with the traditional SATA controller, the read and write performance is greatly improved. The test results show that the designed PCIE to SATA high-speed storage circuit has excellent read and write performance, and has good stability and scalability.

Abstract： CMOS Ring Oscillator has many advantages such as small layout area, large frequency range and easy integration. It is widely used in the DC-DC buck converter and electronic communication systems. On the basis of the traditional ring oscillator circuit, an independent charge and discharge control path is designed to realize a cross charge and discharge ring oscillator circuit, and a lower frequency oscillation period is obtained by externally connecting an off-chip capacitor. Based on 0.18 μm process, the circuit has been simulated with HSIM tool. After the back-end physical implementation, the layout of the oscillation occupies an area of 172 μm×76 μm. Using tools to extract parasitic parameters from the layout and conduct post-simulation, the simulation results indicates that with externally connecting an off-chip capacitor, and the power supply voltage in 3.3 V and temperature in 25℃, the oscillator has a clock period of 1.2 ms. With the power supply voltage changes in 2.7 V~5.5 V and temperature changes in -55℃~125℃, the frequency deviation is 5.83%.The circuit has been successfully applied in a power management chip.

Abstract： Aiming at the problem of low accuracy of high voltage line loss prediction, a line loss prediction model is proposed based on improved BP neural network and Equalization optimizer (EO) algorithm. Firstly, in order to improve the optimization ability of EO algorithm, a variety of chaotic mapping relations is used to initialize the population to increase the population diversity, then the global search ability could be improved. At the same time, the EO algorithm is improved by using the natural selection probability jump strategy, so that the model could jump out of the local optimization according to the probability and converge to the global optimal solution. Secondly, the improved EO algorithm is used to optimize the weight and bias of BP neural network, and the prediction effect of BP neural network for high voltage line loss is improved. Finally, the experimental results show that the proposed line loss prediction model has the highest prediction accuracy compared with regression model, BP neural network model, simulated annealing optimized BP neural network model and EO optimized BP neural network model.

Abstract： The informatization and intelligence of wind farm lock and key management is the key to further improve the level of wind farm operation and maintenance management. This paper designs a smart key security management and control architecture for wind farms. Combined with the proposed short-range wireless communication anti-jamming technology, key encryption, and dynamic attack detection technology, a set of wind farm smart key security management and control system has been developed. The system can realize the functions of opening and closing information traceability and risk prevention, etc., which improves the network, digital and intelligent level of wind farm operation and maintenance.

Abstract：LTE network adopts non orthogonal multiple access to enhance network capacity and D2D to improve spectral efficiency. Allowing D2D users to share the spectrum resources of cellular users using non orthogonal multiple access technology can alleviate the current shortage of spectrum resources, but it will also introduce interference between D2D users and cellular users. This paper proposes a research scheme for the sharing between D2D users and cellular users based on non orthogonal multiple access (NOMA) technology. SIC technology is used to eliminate the interference between cellular users and D2D users. This scheme not only optimizes the system throughput by adjusting the transmission power of users, but also improves the performance of edge users. The simulation results show that the communication scheme between D2D users and cellular users based on NOMA is superior to the traditional cellular resource allocation scheme and interference control scheme in performance.

Abstract：Broadband micro-power (BMP) wireless communication system is a power wireless private network system researched and developed on unlicensed spectrum resources to meet the communication needs of smart grid systems. Although the introduction of spectrum sensing technology of cognitive radio (CR) has alleviated the current problem of shortage of spectrum resources in smart grids, the sensing accuracy of order users under low signal-to-noise ratio still needs to be improved. To solve the above problems, this paper proposes a Variable Step Size Least Mean Square (VSS-LMS) spectrum sensing algorithm based on variable step size. When estimating the amplitude of the originally transmitted signal, the algorithm adds two parameters as a regulator, which can flexibly adjust the step size and smooth the error signal, to solve the contradiction between convergence speed and steady-state error and realize efficient spectrum sensing. Simulation results show that the proposed algorithm not only has good convergence, but also has good detection performance under low signal-to-noise ratio. Compared with the energy detection algorithm and LMS spectrum sensing algorithm, the spectrum detection performance of the system is improved.

Abstract： Meteorological forecast numerical model is an important basis of numerical weather forecasting. The research and development of meteorological numerical model needs to continuously carry out simulation experiments in high-performance computing to test and evaluate the prediction effect. Aiming at the problems of inconvenient, time-consuming, invisible, complex details,difficult to analysis during meteorological scientists manually compile batch processing scripts to carry out numerical simulation experiments, this paper develops an interactive visual model test-bed system based on Python and workflow technology with client-server architecture, which provides the construction, simulation, monitoring, analysis, management and sharing for model experiment. The application results show that, the system improves the efficiency of model research and development and the usability of high-performance computer system, plays an important supporting role in the test-bed of numerical weather forecast model, and has good scalability.

Abstract： In view of the diverse edge computing requirements of the 6G era, reconfigurable technology based on FPGAs can achieve lower latency and provide diversified services. Based on the idea of local dynamic reconfiguration, the ICAP interface is used to reconfigure FPGA resources, so as to realize the local dynamic reconfigurable scheme on the FPGA logic. Drawing on the idea of software process management in the operating system, based on the concept of introducing hardware processes in the Linux operating system, it is possible to divide a whole block of FPGA resources into multiple small FPGA resource blocks, each small reconfigurable FPGA resource block can be abstracted into a hardware process, the hardware process is actually not running on the CPU but running in the FPGA logical resource area, and is only a software language description of the hardware process on the operating system. As a result, the hardware scheme of CPU plus FPGA is designed to achieve partial reconfigurable system, and verified on Xilinx Zynq series chips, and the FPGA hardware resources are scheduled and allocated in a process manner, which greatly improves the utilization and flexibility of FPGA hardware resources.

Abstract：Aiming at the problems that the data visualization analysis of Computational Fluid Dynamics (CFD) software requires a large amount of calculation and the operation is inflexible, this paper implements a CFD data visualization analysis service prototype software for the needs of CFD data visualization analysis. In this prototype software, the adaptive Quality of Service (QoS) optimization technology of the adaptive client is studied to reduce the impact of different client computing environments on the quality of service. The service integration interface based on message middleware is studied and designed to improve the flexibility of service usage. The CFD data visualization analysis service oriented to the cloud computing environment takes advantage of the performance advantages of the cloud server, applies the new development model of "cloud + terminal", and has flexibility and scalability.

Abstract：Aiming at the problem that the OpenFOAM solving software lacks GUI function, the user interaction experience is poor, this paper studies and designs an OpenFOAM solving client software. The paper designs a service-based "cloud + terminal" computing architecture, designs and develops an OpenFOAM solver service, which can be remotely called and accessed by the GUI client, and integrates the visualization service to realize the data visualization analysis function and realize the cloud collaborative solution calculation function. This paper researches and designs a dynamic GUI interface generation method based on interface template, and realizes the user-defined configuration function of the client software interface. The software supports service-based function extension integration,and supports user interface customization and solver user setting interface customization functions. The user interface is friendly and the expansibility is strong. Through the actual calculation example, the verification shows that the client software can realize the complete solution calculation process by calling the cloud solution service and visualization service remotely.

Abstract：With the rapid development of embedded technology and network communication technology, IoT has also developed greatly in recent years, and a large number of different types of end devices need to be connected to the network. However, the whole IoT structure from physical layer to network layer to application layer has different communication protocols between each layer, which greatly increases the difficulty of implementing an integrated system. This paper proposes a design scheme to support custom protocols and multiple end devices for IoT access, and develop a software system accordingly. Accessing different devices through custom protocols enhances the scalability and interoperability of the system, and allows flexible handling of device data and events.

Abstract： Clustering is one of the most important data mining technologies, and K-means is the most famous and commonly used clustering algorithm. However, the performance of K-means depends heavily on the initial centers. It is very important for K-means to select how many initial centers and which data points to choose as the initial centers. Therefore, an initial centers selection method called DPCC (density peak clustering centers) is proposed. DPCC generates a selection decision graph based on density and distance, so as to highlight all density peak points in dataset. These density peak points are the initial centers provided by DPCC for K-means. Experiments show that DPCC not only provides decision support for the number of initial centers, but also improves the accuracy of K-means and reduces the running time of K-means.

Abstract： To avoid the operations of opening embedded device, configuring GPIO pin voltage level, or developing the special upper computer software during the online upgrade process of application program for TMS320F28335, a second bootloader method was proposed. To realize the online upgrade function of application program, TMS320F28335 was configured as Flash boot mode, and the second bootloader was pre-burned. The application program binary file from general serial port upper computer software was stored in the external extended SRAM of TMS320F28335, and was downloaded to on-chip Flash. And then, the system jumps to execute the application program. The experimental results show that this method is effective and feasible, and it has strong universality and reliability.

Abstract：With the development of multi electrified and intelligent aircraft system, the high integration, the system architecture and interface design of airborne electromechanical computer are becoming more and more complex, the development cycle cost is increasing, and the quality problems are becoming increasingly prominent. In view of the severe quality form of airborne electromechanical computer, a model-based system engineering method is proposed to solve complex problems in this paper, which describes the behavior process of system development in a model and standard graphical way. This paper realizes the localization of the MBSE in the development process of airborne electromechanical computer, which improves the quality by putting de verification link in front and can be used as a reference for other complex computer systems.

Abstract： Compared with traditional radio frequency communication, visible light has rich spectrum resources and high security. The multi-channel visible light communication system designed in this paper is divided into transmitter and receiver. Based on Verilog HDL language, the channel distribution module, transparent framing module, channel code block segmentation module at the transmitter, transparent framing module and channel merging module at the receiver are designed in detail, and a multi-channel visible light communication system that supports 16 channels of data at the same time is realized. In order to meet the demand of the high-speed communication of the system, the transmitter / receiver circuit of visible light communication is designed and optimized, and the pre emphasis circuit is used at the transmitter to effectively compensate the channel loss. In the receiver, a T-shaped amplification circuit composed of resistance and capacitance is used to compensate the attenuated signal. Finally, the test is carried out on the experimental platform. The results show that the transmission error rate of the multi-channel visible light communication system designed in this paper is not greater than 10-9, and the single channel transmission rate is not less than 30 Mb/s. This research is conducive to improving the reliability and high speed of the visible light communication system.